Evidence from basic research and from preliminary studies in humans suggests that conditioning factors play an important role in narcotic addiction. After repeated reinforcement, the self-infection ritual may acquire reward properties of its own. A comprehensive treatment program should, therefore, be concerned with conditioned phenomena as with traditional social and psychological rehabilitation. We propose to use the long-acting narcotic antagonist, naltrexone to develop such a behavioral approach. Treatment will consist of either naltrexone alone or naltrexone plus extinction trials. Patients receiving extinction trials will be permitted to self-inject narcotics under naturalistic conditions but the pharmacological effects of the narcotic will be blocked by naltrexone. Patients receiving extinction trials will be compared with controls receiving other treatments (naltrexone without extinction trials, methadone maintenance, and therapeutic community). In addition to the usual outcome measures (drug use, employment, arrests, etc.) we will determine the changes in response to drug-related stimuli. Stimuli will consist of slides, movies, video tapes and actual objects. Responses will be measured by observer, subject, and psychophysiological ratings.